The University of Maine DigitalCommons@UMaine
Earth Science Faculty Scholarship Earth Sciences
12-1-2009 The rC ystal Chemistry of Holtite L. A. Groat
Edward S. Grew University of Maine - Main, [email protected]
R. J. Evans
A. Pieczka
T. S. Ercit
Follow this and additional works at: https://digitalcommons.library.umaine.edu/ers_facpub Part of the Earth Sciences Commons
Repository Citation Groat, L. A.; Grew, Edward S.; Evans, R. J.; Pieczka, A.; and Ercit, T. S., "The rC ystal Chemistry of Holtite" (2009). Earth Science Faculty Scholarship. 100. https://digitalcommons.library.umaine.edu/ers_facpub/100
This Article is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Earth Science Faculty Scholarship by an authorized administrator of DigitalCommons@UMaine. For more information, please contact [email protected]. Mineralogical Magazine,December 2009,Vol. 73(6),pp. 1033–1050
The crystal chemistry of holtite
1, 2 1 3 4 L. A. GROAT *, E. S. GREW ,R.J.EVANS ,A.PIECZKA AND T. S. ERCIT 1 Department of Earth and Ocean Sciences, University of British Columbia, 6339 Stores Road, Vancouver, British Columbia V6T 1Z4, Canada 2 Department of Earth Sciences, University of Maine, 5790 Bryand Global Sciences Center, Orono, Maine 04469-5790, USA 3 Department of Mineralogy, Petrography, and Geochemistry, AGH-University of Science and Technology, Mickiewicza 30, 30-059 Krako´w, Poland 4 Canadian Museum of Nature, Research Division, Ottawa, Ontario K1P 6P4, Canada [Received 26 July 2009; Accepted 22 December 2009]
ABSTRACT
3+ 3+ Holtite, approximately (Al,Ta,&)Al6(BO3)(Si,Sb ,As )S3O12(O,OH,&)S3, is a member of the dumortierite group that has beenfoundinpegmatite, or alluvial deposits derived from pegmatite, at three localities: Greenbushes, Western Australia; Voron’i Tundry, Kola Peninsula, Russia; and Szklary, Lower Silesia, Poland. Holtite can contain >30 wt.% Sb2O3,As2O3,Ta2O5,Nb2O5, and TiO2 (taken together), but none of these constituents is dominant at a crystallographic site, which raises the question whether this mineral is distinct from dumortierite. The crystal structures of four samples from the three localities have beenrefinedto R1 = 0.02À0.05. The results show dominantly: Al, Ta, and vacancies at the Al(1) position; Al and vacancies at the Al(2), (3) and (4) sites; Si and vacancies at the Si positions; and Sb, As and vacancies at the Sb sites for both Sb-poor (holtite I) and Sb-rich (holtite II) specimens. Although charge-balance calculations based on our single-crystal structure refinements suggest that essentially no water is present, Fourier transform infrared spectra confirm that some OH is present in the three samples that could be measured. By analogy with dumortierite, the largest peak at 3505À3490 cmÀ1 is identified with OH at the O(2) and O(7) positions. The single-crystal X-ray refinements and FTIR results suggest the following general formula for holtite: Al7À[5x+y+z]/3 (Ta,Nb)x&[2x+y+z]/3BSi3Ày(Sb,As)yO18ÀyÀz(OH)z, where x is the total number of pentavalent cations, y is the total amount of Sb + As, and z 4 y is the total amount of OH. Comparison with the electron microprobe compositions suggests the following approximate general formulae Al5.83(Ta,Nb)0.50&0.67BSi2.50(Sb,As)0.50O17.00(OH)0.50 and Al5.92(Ta,Nb)0.25&0.83BSi2.00(Sb,As)1.00 O16.00(OH)1.00 for holtite I and holtite II respectively. However, the crystal structure refinements do not indicate a fundamental difference in cation ordering that might serve as a criterion for recognizing the two holtites as distinct species, and anion compositions are also not sufficiently different. Moreover, available analyses suggest the possibility of a continuum in the Si/(Sb + As) ratio between holtite I and dumortierite, and at least a partial continuum between holtite I and holtite II. We recommend that use of the terms holtite I and holtite II be discontinued.
KEYWORDS: holtite, dumortierite, crystal chemistry, discontinued, Western Australia, Kola Peninsula, Lower Silesia.
Introduction the dumortierite group that has beenfoundin HOLTITE, approximately (Al,Ta,&)Al6(BO3) pegmatite, or alluvial deposits derived from 3+ 3+ (Si,Sb ,As )S3O12(O,OH,&)S3, is a member of pegmatite, at three localities: Greenbushes, WesternAustralia (Pryce, 1971; Hoskins et al., 1989); Voron’i Tundry, Kola Peninsula, Russia * E-mail: [email protected] (Voloshin et al., 1977, 1987; Voloshinand DOI: 10.1180/minmag.2009.073.6.1033 Pakhomovskiy 1988); and Szklary, Lower
# 2009 The Mineralogical Society L. A. GROAT ET AL.
Silesia, Poland (Pieczka and Marszałek, 1996). In Voron’i Tundry holtite was reported to contain additionto several distinctivestructural features, more SiO2 (26.6À26.7 wt.% vs. 20.3 wt.% by wet such as face-sharing Al-octahedra and both cation chemistry) thanthe type material, but onlya third and anion vacancies, holtite and dumortierite are of the Sb (5.8À6.4 wt.% vs. 18.0 wt.% as Sb2O3), unique among aluminosilicate minerals because i.e. it was intermediate in composition between they incorporate substantial amounts of the trace type holtite and dumortierite. Moreover, Voloshin elements Nb, Ta, As and Sb, which generally et al. (1977) reported 2.5À3.2 wt.% As2O3 and a form oxides, arsenates or sulphides rather than B2O3 content (4.7À5.1 wt.%) approaching the silicates inpegmatites andother deposits where ideal value consistent with the dumortierite such rare elements are concentrated. Unlike the structure; cf. 1.8 wt.% inthe type holtite, which situation for many new minerals described in the Hoskins et al. (1989) suggested was inerror as the last 50 years, none of the four constituents that structure refinement gave stoichiometric B. In a distinguish holtite from dumortierite is dominant more detailed study of the Voron’i Tundry at a specific crystallographic site, i.e. Si is material, Voloshin et al. (1987) recognized two dominant over Sb3+ and As3+ at the two varieties of holtite: low-Sb holtite I, anearly- tetrahedral sites and Al is dominant over Ta, Nb formed phase, which was described in1977, and and vacancy at the Al(1) site in both minerals. high-Sb holtite II, a late-formed phase. The latter Moreover, the term holtite has beenapplied to an had SiO2 and Sb (as Sb2O3) contents close to ever-widening compositional range that blurred those of the type material; it also contained As, its distinction from dumortierite, which in turn has which Voloshin et al. (1987) also found in a recently been found to contain significant Nb, Ta, sample from Greenbushes. As, Sb and Bi, further muddying the distinction Pieczka and Marszałek (1996) described a third (Groat et al., 2001; Cempı´rek and Nova´k, 2004; holtite occurrenceina pegmatite inSzklary, Borghi et al., 2004; Vaggelli et al., 2004). Lower Silesia, Poland. Containing 12.9 wt.% Sb The primary objective of the present study is to (as Sb2O3), this holtite appeared to be yet another obtain a better understanding of the structural intermediate, but when As is added to Sb (see features of holtite and the crystallographic role below), this holtite is close to holtite II and type played by Nb, Ta, As, Sb and Bi in both holtite holtite inSi and(Sb 3++As3+) content. However, and dumortierite. A secondary objective is to its Ta2O5 content of 5.1 wt.% is but half that of consider the implications of our findings for the other three holtites, i.e. 9.8À14.5 wt.% clarifying the distinction between holtite and (Pryce, 1971; Voloshin et al., 1977, 1987; dumortierite. VoloshinandPakhomovskiy 1988). Groat et al. (2002) found both varieties of holtite in samples from Greenbushes, leaving Brief history of holtite Szklary the only locality where but one variety Holtite was first identified by Pryce (1971) in had been found. Kazantsev et al. (2005) specimens from an alluvial tin deposit, derived concluded that the absence of any intermediate from the weathering of pegmatite, near the compositions allows one to consider the two township of Greenbushes, Western Australia varieties as independent with the main difference (Pryce and Chester, 1978). Weissenberg photo- betweenthe two beingthe Si:(Sb+As) ratio: graphs of single crystals gave space group Pnma, 8.5:1.5 in holtite I and 7:3 in holtite II, and by the same as dumortierite, and indicated the extension, 10:0 in dumortierite. However, T.S. mineral had a crystal structure like dumortierite. Ercit contended in Locock et al. (2006) that the However, chemical analyses showed, that in terms holtite I and holtite II are paragenetic- additionto major Al 2O3 and SiO2 and subordinate compositional varietal names for holtite with B2O3, holtite contains 18.0 wt.% Sb as Sb2O3 inferred, but not proven, structural differences. (originally reported as 4.61 wt.% Sb2O5 and Ercit also pointed out that the varietal distinction 13.89 wt.% Sb2O3), 11.24 wt.% Ta2O5 and is based on amounts of Sb and As subordinate to 0.76 wt.% Nb2O5, constituents not reported in the Si they replace, and that this terminology has dumortierite at that time (e.g. Grew, 1996). not been approved by the Commission on New Voloshin et al. (1977) described a second Minerals, Nomenclature and Classification of the occurrence of holtite in a pegmatite at Mount International Mineralogical Association Vasin-Myl’k, Voron’i Tundry, Kola Peninsula, (CNMNC IMA). Insummary, oneis left with Russia (specific locality from Pekov, 1998). The the impressionthat the nameholtite has become a
1034 THE CRYSTAL CHEMISTRY OF HOLTITE
‘catch-all’ term for pegmatitic dumortierite-like dumortierite, which was described by minerals containing significant Sb, As, Ta and Nb, Golovastikov (1965) and Moore and Araki rather than a mineral with a well defined (1978) as a designonthe semi-regular planar composition. net {6·4·3·4}. Moore and Araki (1978) showed In claiming that Sb is absent in dumortierite, that the net can be broken down into four regions: Kazantsev et al. (2005) was ignoring recent (1) [AlO3] chains of face-sharing octahedra (the chemical data onthis mineral. Groat et al. Al(1) sites) with circumjacent ‘pinwheels’ of six (2001) reported up to 1.0 wt.% Sb2O3 in SiO4 tetrahedra, two Si(1) and four Si(2) sites; dumortierite from localities worldwide. Borghi (2) [Al4O12] cubic close-packed chains, et al. (2004) and Vaggelli et al. (2004) analysed containing the Al(2) and Al(3) octahedral sites, zoned dumortierite in quartzites from that are joined to equivalent chains by reflection Mozambique inwhich Sb 2O3 ranged from 0 to at the O(1) corners of the Al(2) octahedra to form 4.5 wt.% and varied inversely with Si. The Sb- [Al4O11] sheets oriented parallel to (010); rich cores are closer in Sb content to holtite I than (3) [Al4O12] double chains containing the Al(4) to Sb-free dumortierite, and the zoning suggests octahedral sites; the possibility of complete solid solutionbetween (4) BO3 triangles (Fig. 1). holtite I and dumortierite. That is, holtite I might The Al(1)ÀAl(1) distance is ~2.35 A˚ , which is be better classed as Sb-Ta-As rich dumortierite, unusually short for face-sharing octahedra, and leaving open the question whether holtite II is a the Al(1) site is onaverage 75% occupied (Moore species distinct from dumortierite. and Araki, 1978). The Al(1) face-sharing chains are disordered, which results inanaverage chain length that can be adjusted to fit the repeat Previous structure studies distance of the remaining octahedral framework Hoskins et al. (1989) solved and refined the inthe structure (Moore andAraki, 1978). crystal structure of a Greenbushes crystal to R = Hoskins et al. (1989) showed that the crystal 0.030 inspace group Pnma. They determined that structure of holtite differs from that of dumor- the crystal structure is closely related to that of tierite inseveral importantrespects, all of which
FIG. 1. The structure of dumortiertite observed along the a axis.
1035 L. A. GROAT ET AL.
lie withinthe first regionof Moore andAraki structure in(Sb,As)O 3 pyramids, substituting for (1978), i.e. within six-sided tunnels bounded by SiO4 tetrahedra just as SbO3 pyramids substitute the two regions composed of [Al4O12] chains. for SiO4 tetrahedra inholtite from Greenbushes, Both SiO4 tetrahedra are partially replaced by and that there was no preference for either of the 3+ Sb O3 triangular pyramids with no evidence of two Si positions. Kazantsev et al. (2006) derived preference of Sb for one or other of the Si sites, the following chemical formula from their crystal and with Ta replacing Al at the Al(1) position structure study: (Al0.61Ta0.25&)(Al0.96&)2 (Fig. 2). As a result, there are vacancies at the (Al0.96&)2 (Al0.90&)2 (Si2.49Sb0.35As0.13) coordinating anion sites [O(2) and O(7)] as well O13.46(O0.48OH0.52)(BO3). as at the Al(1) site. Relative to the Si positions, Using the Rietveld method, Zubkova et al. the Sb3+ sites are shifted about 0.5 A˚ closer to the (2006) refined the crystal structure of a Voron’i Al(1) positionto accommodate the longer Tundry holtite II from powder data (to Rp = Sb3+Àanion bonds (average ~1.9 A˚ ). Whenthe 2.93% and Rb = 2.10%) and found that the crystal Sb sites are occupied, the adjacent Si, O(2) [for structure differs from those of holtite I and of the Si(1)], and O(7) [for Si(2)] positions are vacant. Greenbushes holtite (Hoskins et al., 1989), which Hoskins et al. (1989) obtained the formula is also holtite II, inthat the Si(1) positionis (Si2.25Sb0.75)B[Al6 (Al0.43Ta0.27& 0.30) vacant, the Sb(1) position is only partially O15(O,OH)2.25](Z = 4) from their crystal- occupied and approximately 5% of the Al atoms structure study. at the Al(2) site are replaced by Sb5+ cations, 5+ Both Pryce (1971) and Hoskins et al. (1989) giving (Ta0.30Al0.26&)(Al0.95Sb0.05)2Al2 3+ 3+ reported diffuse layers normal to a at spacings of (Al0.98&)2(Si0.65Sb0.30As0.05)3(Sb0.44&)O9.30 2a and 3a inrotationphotographs of holtite. (O,OH)4.56(BO3). The refined Si content is about Hoskins et al. (1989) suggested that the diffuse two thirds of that measured by electronmicro- layers visible inthe photographs indicatedthat probe, a discrepancy that Zubkova et al. (2006) although there is an ordering effect in any one attributed to contamination by quartz and individual tunnel, it is not strongly correlated with amorphous silica encasing the studied holtite that of adjacent tunnels. They went on to suggest fibres. However, the measured SiO2 is that a complete replacement of SiO4 rings by 17.31 wt.% (Kazantsev et al., 2005), a value SbO3 rings takes place at ordered intervals, which typical of holtite II. are likely related to the order of the vacancies at the Al(1) site. Kazantsev et al. (2005, 2006) refined the Experimental crystal structure of anAs-bearingholtite I Preliminary examination of samples from all three crystal from Voron’i Tundry to R = 0.046. They localities with a Philips XL30 scanning electron reported that As3+ is incorporated into the crystal microscope equipped with anenergy-dispersion
FIG. 2. Dispositionof SiO 4 tetrahedra (left) and (Sb,As)O3 groups (right) and the coordinated central Al(1) site in holtite (after Hoskins et al., 1989).
1036 THE CRYSTAL CHEMISTRY OF HOLTITE
X-ray spectrometer revealed holtite I inthe (taking into account the different multiplicities of sample from Voron’i Tundry (labelled H1), the sites). holtite II inthe specimenfrom Szklary (H4), (2) A cationwas assumed to occupy the Al(1) site and both holtite I and holtite II in the sample from when there were oxygen atoms at the coordinating Greenbushes (H2 and H3). O(2) and O(7) positions. Therefore, the atomic For the X-ray experiments, crystals of samples occupancy of the Al(1) site was fixed at the value H1À3 were ground to approximate spheres with a obtained through refinement for the O(2) and O(7) Nonius sphere grinder. This was not possible for (and hence Si(1) and Si(2)) sites. H4, which has a fibrous habit, so a single fibre (3) The ratio of Al:Ta was refined within this was carefully extracted from the mass. Single- overall fixed value. crystal X-ray diffractionmeasurementswere (4) The Sb(1) site was assumed to be coordinated made at C-HORSE (the Centre for Higher Order by three O atoms in trigonal coordination, and to Structure Elucidation, in the Department of be occupied whenthe Si(1) site was empty; the Chemistry at UBC) using a Bruker X8 APEX II same assumptions were made for the Sb(2) and diffractometer with graphite monochromated Mo- Si(2) sites. The overall occupancy of the Sb(1) Ka radiation. Data were collected in a series of f and Sb(2) sites was fixed at values inverse of and o scans in 0.50º oscillations with 10.0 second those refined for the Si(1) and Si(2) [and O(2) and exposures. The crystal-to-detector distance was O(7)] sites. 40 mm. Data were collected and integrated using (5) The ratio of Sb:As was refined within this the Bruker SAINT software package. Data for overall fixed value. samples H1À3 were corrected for absorption (6) When necessary, the overall occupancies of effects using the multi-scan technique (SADABS) the Al(1), Sb(1) and Sb(2) sites were adjusted to and were corrected for Lorentz and polarization match the refined occupancies of the Si(1), Si(2), effects. Examination of the data for the H4 O(2) and O(7) sites. samples showed that the crystal is made up of (7) The occupancies of the Al(2), Al(3) and Al(4) three twinindividualsrelated by 120º rotation sites were refined. about a threefold twinaxis parallel to a. The data Moore and Araki (1978) and Alexander et al. were resolved using CELL NOW and were (1986) suggested that the apparent partial corrected for absorptioneffects using TWINABS occupancies of the Al(2), Al(3) and Al(4) sites and then for Lorentz and polarization effects. indumortierite were due to a correlationbetween All refinements were performed using the site occupancies and thermal motion. Alexander SHELXTL crystallographic software package of et al. (1986) therefore scaled the refined Bruker AXS. The structures were refined using occupancies of the Al(2À4) sites by dividing starting parameters from Hoskins et al. (1989). them by 0.95. However, Ferraris et al. (1995) Scattering factors for neutral atoms were used for suggested that for their refinement of the crystal the cations, and ionic factors for O2À for oxygen. structure of magnesiodumortierite there was no All atoms were refined anisotropically. The valid reasonto disregard the refinedvalues. Inour weighting scheme was based on counting study, constraining these sites to be fully occupied statistics. Neutral atom scattering factors were with Al resulted inincreasesinthe R values by at takenfrom Cromer andWaber (1974). least 1%, and attempts to refine for cations in Anomalous dispersion effects were included in additionto Al (such as Ti) were unsuccessful. Fcalc. (Ibers and Hamilton, 1964); the values for After collectionof X-ray diffractiondata, the Df ’ and Df ’’ were those of Creagh and McAuley single crystals were attached to Lucite disks with (1992). The values for the mass attenuation Petropoxy and polished for the electron micro- coefficients are those of Creagh and Hubbell probe study. Inorder to obtainthe most accurate (1992). The following assumptions were made possible compositions, electron microprobe data and appropriate constraints applied: were collected from all of the crystals (except H2) (1) The Si(1) and Si(2) sites were assumed to using two different instruments, one at University contain only silicon atoms or be vacant. When of British Columbia (UBC) and the other at the occupied, the sites were assumed to be coordi- Canadian Museum of Nature (CMN). nated by four O atoms in tetrahedral coordination. At UBC, compositions were obtained with a The O(2) and O(7) sites were assumed to contain fully-automated Cameca SX50 microprobe, oper- only oxygen atoms or be vacant. The occupancies ating in the wavelength-dispersion mode with the of all four sites were constrained to be equivalent following operating conditions: excitation
1037 L. A. GROAT ET AL.
voltage, 20 kV; probe current, 20 nA; peak count reductionwas carried out with the ‘‘PAP’’ f(rZ) time, 20 s; background count time, 10 s; final method (Pouchou and Pichoir, 1985) and was beam diameter, 10 mm. Data reductionwas checked with CITZAF (Armstrong, 1995). For the carried out with the ‘‘PAP’’ f(rZ) method elements considered, the following standards and (Pouchou and Pichoir, 1985). For the elements X-ray lines were used: kyanite, Al-Ka, Si-Ka; considered, the following standards and X-ray apatite, P-Ka; rutile, Ti-Ka; hematite, Fe-Ka; lines were used: kyanite, Al-Ka, Si-Ka; apatite, GaAs, As-La; stibiotantalite, Nb-La,Sb-La, P-Ka; rutile, Ti-Ka; synthetic fayalite, Fe-Ka; Ta-Ma. tennantite, As-Ka; columbite, Nb-La; tetrahedrite, Formulae were calculated from the electron Sb-La; microlite, Ta-Ma. microprobe data using the number of oxygen At CMN, compositions were obtained with a atoms determined from the X-ray refinements and JEOL 733 electronmicroprobe with Geller assuming 1 B atom per formula unit (a.p.f.u.). automation, operating in the wavelength-disper- Fourier transform infrared spectroscopy (FTIR) sion mode with the following operating condi- was performed at the Canadian Conservation tions: excitation voltage, 15 kV; probe current, Institute on a holtite sample from Szklary (H4) 20 nA; peak count time, 50 s; background count and two from Greenbushes, one obtained from the time, 25 s; final beam diameter, 10 mm. Data AustralianMuseum of Mineralogy (sample
TABLE 1. Average electronmicroprobe compositionsof holtite.
Sample H1 H2 H3 H4 Holtite I Holtite I Holtite II Holtite II n =11 n =11 n =2 n =10
Wt.% P2O5 0.19 0.18 0.00 0.24 Nb2O5 0.21 0.18 0.05 1.12 Ta2O5 12.06 13.15 10.93 5.06 SiO2 22.31 23.66 18.24 18.02 TiO2 0.01 0.00 0.09 1.46 B2O3 5.12 5.20 4.98 5.20 Al2O3 46.20 46.85 43.50 47.62 Fe2O3 0.00 0.01 0.32 0.22 As2O3 3.35 2.33 2.80 3.86 Sb2O3 6.81 4.85 17.29 15.04 Total 96.26 96.41 98.20 97.84 P5+ 0.018 0.017 0.000 0.023 Nb5+ 0.011 0.009 0.003 0.056 Ta5+ 0.371 0.399 0.345 0.153 Si4+ 2.524 2.638 2.120 2.008 Ti4+ 0.001 0.000 0.008 0.122 B3+ 1111 Al3+ 6.160 6.156 5.959 6.255 Fe3+ 0.000 0.001 0.028 0.018 As3+ 0.230 0.158 0.198 0.261 Sb3+ 0.318 0.223 0.828 0.691 As + Sb 0.548 0.381 1.026 0.952 Nb + Ta 0.382 0.408 0.348 0.209 O 17.610 17.643 17.145 17.180
Number of analyses used in average = n. Mg sought but not detected. Formulae were calculated using the number of oxygen atoms determined from the X-ray refinements and assuming 1 B a.p.f.u.
1038 THE CRYSTAL CHEMISTRY OF HOLTITE
D43778, labelled A) and the other from the between the very fine fibres of holtite. In addition, Smithsonian Institution (sample NMNH 128674, we suspect that the data reductionroutinesare labelled S). The available amount of Voron’i unable to correct completely for the effect of As, Tundry material was insufficient for FTIR. Each Sb, and Ta on Si and Al, elements not normally sample was separated from the rock and washed analysed together. The results confirm that with ethanol. The sample was positioned on a samples H1 and H2 are holtite I and H3 and H4 Spectra-Tech low-pressure diamond microsample are holtite II, with a maximum of 17.29 wt.% cell and analysed using a Bruker Optics Hyperion Sb2O3 (0.82 Sb a.p.f.u.) inH3. The compositions 2000 microscope interfaced to a Tensor 27 also show 2.33À3.86 wt.% As2O3 (0.16À0.26 As spectrometer. The spectrum was acquired inthe a.p.f.u.) and, as expected, Si+As+Sb equals 4000 to 430 cmÀ1 range by co-adding 150 scans. ~3 a.p.f.u. Samples H1, H2 and H3 contain similar amounts of Ta (0.34À0.40 Ta a.p.f.u.); H4 contains approximately half as much Ta Results (0.15 Ta a.p.f.u.) but considerably more Nb Average electronmicroprobe compositionsof the (0.06 a.p.f.u.) and Ti (0.12 a.p.f.u.). The composi- structure crystals are giveninTable 1. The totals tions also show 43.50À47.62 wt.% Al2O3 are low, evenif water is present.The (sub)micro- (5.88À6.26 Al a.p.f.u.). Three of the samples fibrous character of holtite is a possible cause of also show traces of P (to 0.02 P a.p.f.u.). the low analytical totals. Grew et al. (2008) Data collection and refinement parameters for attributed low totals for analyses of finely porous the single-crystal X-ray experiments are summar- material to contamination from epoxy filling the ized inTable 2, positionalparameters inTable 3 pores; similarly, epoxy could have filled the gaps and bond lengths in Table 4. A full list of atomic
TABLE 2. Data measurement and refinement information.
SpecimenH1 H2 H3 H4 a (A˚ ) 4.6818(4) 4.6981(7) 4.6900(6) 4.7264(9) b (A˚ ) 11.867(1) 11.926(2) 11.897(4) 11.981(2) c (A˚ ) 20.302(2) 20.413(3) 20.385(6) 20.566(4) V (A˚ 3) 1127.9(2) 1143.7(3) 1137.4(5) 1164.6(4) Space Group Pnma Z 4 Crystal size (mm) 0.1760.1460.14 0.1460.1160.08 0.1260.0860.06 0.1860.0260.02 À3 Dcalc. (Mg m ) 4.916 4.848 4.875 4.762 RadiationMo- Ka Monochromator Graphite Total Fo 29307 10309 3871 1215 Unique Fo 1754 1733 1741 1057 Fo >4s Fo 1693 1508 1439 826 Rint 0.03(1) 0.03(2) 0.03(3) À L.s. parameters 165 165 165 167 Range of h À6 4 6 À5 4 604 604 5 Range of k À14 4 16 À15 4 16 0 4 16 À14 4 14 Range of l À25 4 28 À25 4 28 À28 4 28 À24 4 24 R1 for Fo >4s Fo 0.0161 0.0172 0.0214 0.0511 R1 for all unique Fo 0.0170 0.0238 0.0297 0.0795 wR2 0.0411 0.0387 0.0446 0.1398 a 0.0142 0.0202 0.0183 0.0401 b 1.07 0.01 0.01 18.65 GoF (= S) 1.242 1.037 0.963 1.174 ˚ À3 Drmax (e A ) 0.953 0.471 0.861 1.836 ˚ À3 Drmin (e A ) À0.525 À0.326 À0.429 À1.013
2 2 2 2 2 w = 1/[s (Fo)+(a’P) +b’P] where P = [Max (Fo,0)+2’Fc)]/3
1039 L. A. GROAT ET AL.
TABLE 3. Atomic parameters for holtite.
H1 H2 H3 H4
Al(1) x 0.40099(6) 0.40077(6) 0.40563(8) 0.4048(7) y ¯Ù˘ ¯Ù˘ ¯Ù˘ ¯Ù˘ z 0.25024(1) 0.25027(1) 0.24991(2) 0.2504(2) Ueq 0.01537(9) 0.01530(9) 0.0050(1) 0.030(1) nAl 0.2753(5) 0.2799(5) 0.2291(5) 0.310(2) nTa 0.1597(5) 0.1618(5) 0.1304(5) 0.053(2) Al(2) x 0.55777(8) 0.55782(9) 0.5571(1) 0.5577(8) y 0.61051(3) 0.61049(3) 0.61078(5) 0.6107(3) z 0.47331(2) 0.47333(2) 0.47353(3) 0.4733(2) Ueq 0.0054(1) 0.0053(1) 0.0039(2) 0.012(1) n 0.986(3) 0.993(2) 0.994(4) 1.00(2) Al(3) x 0.05959(8) 0.05953(9) 0.0586(1) 0.0594(8) y 0.49029(3) 0.49028(3) 0.48992(5) 0.4893(3) z 0.43152(2) 0.43151(2) 0.43149(3) 0.4311(2) Ueq 0.0053(1) 0.0049(1) 0.0041(2) 0.010(1) n 0.984(3) 0.989(2) 0.992(4) 1.00(2) Al(4) x 0.06075(9) 0.0602(1) 0.0598(1) 0.0596(8) y 0.35985(4) 0.35952(4) 0.35925(5) 0.3577(3) z 0.29022(2) 0.29005(2) 0.29023(3) 0.2894(2) Ueq 0.0079(2) 0.0073(2) 0.0061(2) 0.013(1) n 0.916(3) 0.930(3) 0.971(4) 0.96(2) Si(1) x 0.0868(3) 0.0862(3) 0.0874(4) 0.089(4) y ¯Ù˘ ¯Ù˘ ¯Ù˘ ¯Ù˘ z 0.40833(7) 0.40839(6) 0.4079(1) 0.4080(7) Ueq 0.0054(2) 0.0050(2) 0.0033(3) 0.018(3) n 0.435(2) 0.442(2) 0.359(2) 0.363(5) Sb(1) x 0.1109(6) 0.1103(8) 0.1135(3) 0.113(3) y ¯Ù˘ ¯Ù˘ ¯Ù˘ ¯Ù˘ z 0.3861(2) 0.3865(2) 0.38349(8) 0.3864(4) Ueq 0.0107(6) 0.0103(8) 0.0067(3) 0.016(2) nSb 0.033(2) 0.028(2) 0.120(3) 0.11(1) nAs 0.032(2) 0.030(2) 0.021(3) 0.03(1) Si(2) x 0.5875(1) 0.5871(2) 0.5878(2) 0.591(2) y 0.5205(1) 0.52053(9) 0.5209(2) 0.5225(7) z 0.32972(4) 0.32979(4) 0.32970(6) 0.3295(4) Ueq 0.0055(2) 0.0054(2) 0.0041(2) 0.009(1) N 0.870(4) 0.883(3) 0.719(5) 0.73(1) Sb(2) x 0.6091(3) 0.6086(4) 0.6108(2) 0.610(1) y 0.5596(2) 0.5589(2) 0.5619(1) 0.5596(5) z 0.31736(8) 0.31747(9) 0.31676(4) 0.3171(4) Ueq 0.0089(4) 0.0093(5) 0.0060(2) 0.015(1) nSb 0.086(4) 0.077(3) 0.276(5) 0.20(1) nAs 0.044(4) 0.039(3) 0.005(5) 0.08(1) B x 0.2280(4) 0.2274(5) 0.2265(6) 0.234(4) y ÜÜÜÜ z 0.4156(1) 0.4155(1) 0.4160(2) 0.416(1) Ueq 0.0070(4) 0.0066(4) 0.0047(5) 0.017(3) O(1) x 0.3777(3) 0.3774(3) 0.3790(4) 0.376(2) y ¯Ù˘ ¯Ù˘ ¯Ù˘ ¯Ù˘ z 0.45681(7) 0.45674(7) 0.4573(1) 0.4566(6) Ueq 0.0072(3) 0.0063(3) 0.0057(4) 0.013(2)
10 4 0 THE CRYSTAL CHEMISTRY OF HOLTITE
Table 3 (contd.)
H1 H2 H3 H4
O(2) x 0.1613(4) 0.1613(4) 0.1622(7) 0.159(4) y ¯Ù˘ ¯Ù˘ ¯Ù˘ ¯Ù˘ z 0.33042(8) 0.33048(8) 0.3303(1) 0.3286(9) Ueq 0.0104(4) 0.0106(4) 0.0086(7) 0.022(3) n 0.435(2) 0.442(2) 0.359(2) 0.363(5) O(3) x 0.8943(2) 0.8943(2) 0.8914(3) 0.891(2) y 0.63873(8) 0.63880(8) 0.6380(1) 0.6381(6) z 0.42518(5) 0.42516(5) 0.42427(7) 0.4247(4) Ueq 0.0068(2) 0.0062(2) 0.0055(3) 0.013(2) O(4) x 0.4039(2) 0.4035(2) 0.4017(3) 0.401(2) y 0.43460(8) 0.43452(8) 0.4349(1) 0.4363(6) z 0.28240(5) 0.28251(5) 0.28166(7) 0.2824(4) Ueq 0.0082(2) 0.0073(2) 0.0062(3) 0.012(2) O(5) x 0.3942(2) 0.3944(2) 0.3918(3) 0.392(2) y 0.54999(8) 0.54984(8) 0.5507(1) 0.5513(6) z 0.39419(5) 0.39427(5) 0.39412(6) 0.3948(4) Ueq 0.0069(2) 0.0063(2) 0.0050(3) 0.010(1) O(6) x 0.8811(2) 0.8807(2) 0.8828(3) 0.886(1) y 0.45102(8) 0.45119(8) 0.4506(1) 0.4475(6) z 0.35231(5) 0.35208(5) 0.35194(6) 0.3518(4) Ueq 0.0078(2) 0.0070(2) 0.0060(3) 0.013(1) O(7) x 0.6585(3) 0.6586(3) 0.6607(4) 0.658(2) y 0.6333(1) 0.6333(1) 0.6331(2) 0.635(1) z 0.28894(6) 0.28880(6) 0.28848(9) 0.2880(6) Ueq 0.0116(3) 0.0108(3) 0.0086(6) 0.022(2) n 0.870(4) 0.883(3) 0.719(5) 0.73(1) O(8) x 0.1752(3) 0.1726(3) 0.1737(4) 0.178(3) y ÜÜÜÜ z 0.34978(7) 0.34990(7) 0.35013(9) 0.3498(6) Ueq 0.0100(3) 0.0087(3) 0.0071(4) 0.012(2) O(9) x 0.2539(2) 0.2537(2) 0.2540(3) 0.253(2) y 0.35050(8) 0.35049(8) 0.3506(1) 0.3504(8) z 0.44768(5) 0.44776(5) 0.44791(6) 0.4474(6) Ueq 0.0074(2) 0.0068(2) 0.0057(3) 0.018(2) O(10) x 0.7568(3) 0.7576(3) 0.7611(4) 0.769(2) y ÜÜÜÜ z 0.27390(7) 0.27364(7) 0.2732(1) 0.2730(8) Ueq 0.0109(3) 0.0102(3) 0.0080(4) 0.014(3) O(11) x 0.7501(2) 0.7500(2) 0.7500(3) 0.749(1) y 0.46706(8) 0.46710(8) 0.4673(1) 0.4683(7) z 0.48832(4) 0.48826(5) 0.48840(6) 0.4871(5) Ueq 0.0054(2) 0.0052(2) 0.0037(3) 0.007(2)
displacement parameters and bond angles confirm that the refinement strategy outlined (Tables S1 and S2 respectively), have been above is appropriate. deposited with the Principal Editor of The average AlÀO distances are remarkably Mineralogical Magazine and are available from similar for all four samples, at ~1.99 A˚ for www.minersoc.org/pages/e_journals/
10 41 L. A. GROAT ET AL.
TABLE 4. Interatomic distances (A˚ ).
H1 H2 H3 H4
Al(1)ÀO(2) 1.977(1) 1.987(2) 1.997(3) 1.98(2) Al(1)ÀO(2)a 2.041(2) 2.053(2) 2.030(3) 2.02(2) Al(1)ÀO(7)b,c 62 1.959(1) 1.967(1) 1.966(2) 1.97(1) Al(1)ÀO(7)d 62 1.997(1) 2.006(1) 1.996(2) 1.99(1)
10 42 THE CRYSTAL CHEMISTRY OF HOLTITE
According to Moore and Araki (1978) ‘‘...the for the Si(1) and Si(2) sites, and ~12% Sb+As Al4ÀO10 distance dilations can be easily (6% Sb and 6% As at the Sb(1) site, and 8% Sb explained by noting that all four Al cations and 4% As at the Sb(2) site) and 88% vacancies surround and are practically coplanar to O10 for the Sb(1) and Sb(2) positions. and, therefore, exhibit uniform cation–cation For the holtite II samples, the Si(1) and Si(2) repulsionaway from centralO10... ’’. The distance positions showed refined occupancies of ~72% Si dilations result in bond-valence totals of ~1.5 v.u. and 28% vacancies. The Sb(1) and Sb(2) site to the O(10) position. The Al(1)ÀAl(1) separation occupancies refined to ~28% vacancies, therefore is similar for crystals H1, H2, and H3 the Sb(1) and Sb(2) occupancies are 28% Sb+As (2.341À2.349 A˚ ) but slightly longer for H4 and 72% vacancies, but the relative amounts of Sb (2.363 A˚ ). and As were different for the two samples (24% Sb For the holtite I samples (H1 and H2), the and 4% As at Sb(1) for H3, vs. 22% Sb and 6% As refined occupancies of the Al(1) site were very at Sb(1) for H4, and 28% Sb and 1% As at Sb(2) similar at ~56% Al, 32% Ta and a 12% vacancy. for H3, and 20% Sb and 8% As at Sb(2) for H4). For the holtite II sample H3, the Al(1) site Overall, it is remarkable how little the bond occupancy refined to ~46% Al, 26% Ta, and a lengths and angles of the AlO6 octahedral 28% vacancy. However, the refined Al(1) site framework change in response to major substitu- occupancy for the other holtite II sample (H4) tions, including vacancies, at the Al(1), Si, and Sb refined to 62% Al, 11% Ta, and a 27% vacancy. positions. This probably reflects the presence of Ti and Nb, The FTIR spectra of samples A, S, and H4 in as shownby the electronmicroprobe composi- the 4000À450 cmÀ1 regionare showninFig. 3 a; tions; studies of dumortierite (Alexander et al., a list of absorbance bands is given on the left side 1986; Platonov et al., 2000) have shownthat Ti of Table 5. The spectra are similar to previously substitutes at the Al(1) site, and we would expect reported IR spectra of holtite and dumortierite Nb to do so also, based onits similarity invalence (Voloshin et al., 1977, 1987; Voloshinand and radius to Ta. The occupancies of the Al(2) Pakhomovskiy, 1988; Werding and Schreyer, and Al(3) positions refined to between 98 and 1990; Fuchs et al., 2005). The right side of 100% Al for all of the samples studied. However, Table 5 lists absorbance peaks in holtite I, holtite the occupancy of the Al(4) position was lower for II, and dumortierite from Voloshin et al. (1987) the holtite I samples (92 and 93% Al for H1 and for comparisonwith the currentspectra. H2 respectively) thanfor the holtite II samples The spectra of samples A, S and H4 each show (96 and 97% Al for H3 and H4 respectively). a series of peaks corresponding to OH stretching The average Si(1)ÀO and Si(2)ÀO distances inthe 4000 À3000 cmÀ1 region(shownin are slightly shorter (1.64À1.65 and 1.64 A˚ Fig. 3b), confirming the presence of hydrogen in respectively) for the holtite I samples thanfor the holtite structure (see below). The spectrum of the holtite II crystals (1.66À1.67 and H4 also shows a small peak at 1648 cmÀ1 due to ˚ 1.65À1.68 A), reflecting the difference in Si site bending of non-structural H2O (Voloshin et al., occupancies (88% in holtite I vs. 72% in 1987; VoloshinandPakhomovskiy, 1988; holtite II). Similarly, the average Sb(1)ÀOand Rossman, 1988). Sb(2)ÀO distances are shorter (1.86À1.87 and Voloshin et al. (1987) and Voloshin and 1.90À1.91 A˚ respectively) for the holtite I phases Pakhomovskiy (1988) compared several IR thanfor the holtite II samples (1.89 À1.91 and spectra of holtites with that of dumortierite, and 1.93À1.94 A˚ ). identified certain peaks as characteristic of holtite The Si(1)ÀSb(1) and Si(2)ÀSb(2) separations I and holtite II, appearing in one variety but not are similar for the holtite I crystals (0.46À0.47 the other. The spectra of A and S (holtite I) and and 0.53À0.54 A˚ respectively) and longer for the H4 (holtite II) show both types of peaks. For holtite II crystal H3 (0.51 and 0.57 A˚ ). The example, Voloshin et al. (1987) and Voloshin and twinned holtite II crystal H4 shows a similar Pakhomovskiy (1988) identified peaks at 750 and Si(1)ÀSb(1) separationto the holtite I samples 545 cmÀ1 as characteristic of holtite I, and peaks (0.46 A˚ ) but the shortest Si(2)ÀSb(2) distance at 940À950 cmÀ1 and 825 cmÀ1 as characteristic (0.52 A˚ ). of holtite II; however, all of these peaks appear to For the holtite I samples, the refined occupan- some degree ineach of our three samples. cies of the Si(1), Sb(1), Si(2) and Sb(2) sites were Similarly, a side-peak at 1315 cmÀ1 onthe main À1 also very similar, at ~88% Si and 12% vacancies antisymmetric BO3 stretching peak at 1360 cm
10 43 L. A. GROAT ET AL.
is identified by Voloshin et al. (1987) and VoloshinandPakhomovskiy (1988) as character- istic of holtite I, but only one (sample A) of our two holtite I samples show this feature (Fig. 3c). This suggests that holtite I and holtite II do not have distinct FTIR signatures. The spectra of A, S and H4 each show a group of three small peaks around 2900 cmÀ1 which do not correspond to any peaks in the Voloshin et al. (1987) and Voloshin and Pakhomovskiy (1988) spectra. These may be due to CÀO stretching in residue left behind after the samples were rinsed with ethanol and dried during preparation.
Water in holtite + Pryce (1971) measured 0.38 wt.% H2O and À 0.08 wt.% H2O inthe type specimen,a holtite II, but the questionwhether H 2O is anessential constituent of holtite has remained unresolved. In dumortierite, OH is thought to occur primarily at the O(2) and O(7) positions (Moore and Araki, 1979; Alexander et al., 1986; Werding and Schreyer, 1990; Ferraris et al., 1995; Cempı´rek and Nova´k, 2005; Fuchs et al., 2005). However, Hoskins et al. (1989) wrote that ‘‘...the (occu- pancy) factors obtained for O2 and O7...are in good agreement with the overall metal occupancy in Al1... For this reason there is not the need in holtite (as there is indumortierite) for OH À replacement of O2À to provide a reductioninlocal charge to compensate for...a vacancy in Al1...’’. Nonetheless, there have been several reports of H2O by infrared spectroscopy or single-crystal refinement. Voloshin et al. (1977) reported traces + and 1.13 wt.% H2O in their ‘anhydrous’ and ‘hydrated’ varieties of holtite I from Voron’i Tundry, a distinction confirmed by IR evidence for OH inthe latter, whereas Voloshin et al. (1987) reported IR evidence for H2O (most likely as OH) in both holtite I and holtite II. Finding O(10) undersaturated in their refinement of holtite I from Voron’i Tundry, Kazantsev et al. (2005) concluded it was occupied by both O and OH. Kazantsev et al. (2006) calculated the occupancy to be 0.52 OH in a revision of the refinement.
FIG. 3. FTIR spectra of samples A and S (both from Greenbushes, Western Australia) and H4 (from Szklary, Lower Silesia, Poland): (a) Full spectra; (b) 4000 cmÀ1 to 3000 cmÀ1 region showing OH stretching bands; (c) 1200 cmÀ1 to 1500 cmÀ1 region showing antisym- metric BO3 stretching bands.
10 4 4 THE CRYSTAL CHEMISTRY OF HOLTITE
À1 TABLE 5. Infrared absorption bands of holtite between 4000À450 cm .
A (Australia) S (Australia) H4 (Poland) —— From Voloshin et al. (1987) —— Holtite I Holtite I Holtite II Holtite I Holtite II Dumortierite
3697 3690 3625 3623* 3615 3504 3501 3490 3500À3510 3495À3500 3500 3480 3400À3420 3400 3340 2957 2957* 2954 2925 2925 2919 2854 2854 2850 1727 1648 1630À1640 1650 1650 1415 1415 1415 1415 1415À1420 1415 1362 1363 1363 1370 1360À1370 1375 1312 1315 1170 1165À1175 1170 1128 1126 1121 1120À1125 1090À1095 1090À1095 1090 1045 1007 1006 1006 1030À1020 1015 1015 952 952 950 940À950 920{ 920{ 920À925 893{ 892 890{ 866 868 872 860À870 865À870 860 829 827 827 825 797À800 795 770À775 760À762 777 742 743 744{ 750 762 728 728 727 725 720À725 725 699{ 698 698 695À700 700 636 636 635 635À640 650À652 640 605À610 591 589 592 590À600 585À590 590 558 561{ 560{ 544{ 538{ 538{ 545 535 497{ 497{ 520 500 471 471{ 472 465À470 460À462 470 459 459 458
Peak positions are reported in cmÀ1. Blank lines indicate a peak is absent in the corresponding spectrum. * À very small peak, not visible in full spectrum shown in Fig. 3. { À shoulder
Although charge-balance calculations based on (1978), the positions and integrated area of our single-crystal structure refinements (see individual peaks in an IR spectrum depend on below) suggest that essentially no water is coordination and the local charge environment. present, the FTIR results confirm that some OH Assuming OH is located at the O(2) and O(7) is present in the samples measured. The peaks in positions (each coordinated by one tetrahedral and the OH stretching region (Fig. 3b) closely two octahedral cations), Fuchs et al. (2005) resemble those reported for dumortierite (Fuchs identified the main dumortierite peak at et al., 2005; Werding and Schreyer, 1990). As 3495 cmÀ1 with a nine-charge coordination noted by Rossman (1988) and Povarennykh environment for OH, such as (Si4+,Al3+, [6]M2+)
10 45 L. A. GROAT ET AL.
or ([4]T3+,Al3+,Al3+). This is the largest OH peak of oxygen atoms around the two adjacent Al(1) inthe holtite spectra, andit is located at positions, it is likely that a complete replacement À1 À1 3501 cm and 3504 cm inthe two of SiO4 rings by SbO3 rings takes place, as is Greenbushes samples A and S respectively, and suggested by Hoskins et al. (1989). The structural at 3490 cmÀ1 inH4. If this peak is due to OH at channel in end-member dumortierite can be O(2) and/or O(7), another nine-charge environ- represented as a chain of Al3+ cations alternating ment that must be considered is (Si4+, [6]&,Ta5+). with rings of three SiO2+ groups: Higher cation valences tend to shift absorbance 2+ 3+ 2+ 3+ 2+ ...(3SiO )(Al )(3SiO )(Al )(3SiO )... peaks to higher wavenumbers, so the prevalence of this latter environment may be responsible for A single SiO2+ ring in the channel is replaced the shift to values above 3500 cmÀ1 inthe by anSb 3+ ring as follows: Greenbushes samples. Other peaks identified in 2+ 3+ 2+ 3+ 2+ (3SiO )(Al )(3SiO )(Al )(3SiO ) ? the IR spectrum of dumortierite by Fuchs et al. (3SiO2+)(&)(3Sb3+)(&)(3SiOH3+) (2) (2005) are the eight-charge environment such as (Si4+, [6]M2+, [6]M2+) or (Si4+,Ti4+, [6]&) around This results inloss of coodinationtothe two 3620 cmÀ1, the seven-charge environment (Si4+, adjacent Al(1) sites, which must now remain Al3+, [6]&) at 3675 cmÀ1, and the six-charge vacant. Hydrogen ions are added to oxygen atoms environment (Si4+, [6]M2+, [6]&) at 3696 cmÀ1. at three of the six coordinating O(2) and O(7) sites Spectra for samples A and H4 show small peaks inorder to maintaincharge balance. near 3625 cmÀ1 which may correspond to the Substitution 2 represents the insertion of a short eight-charge environment, and sample A shows a &ÀSb segment into the AlÀSiO chain. The peak at 3697 cmÀ1 which may correspond to the &ÀSb segment can be expanded to preserve six-charge environment. The seven-charge envir- octahedral coordination for Al3+ (or other cations) onment peak is absent from all three spectra. by the following additional substitution: (&)(3SiOH3+)(Al3+)(3SiO2+) ? 3+ 3+ Chemical substitution and ordering in the (&)(3Sb )(&)(3SiOH ) (3) holtite structure A combination of substitution 2 followed by Precessionphotographs synthesizedfrom the many substitutions 3 can create an arbitrarily long X-ray data show diffuse layers mentioned by &ÀSb segment while introducing just three OHÀ Pryce (1971) and Hoskins et al. (1989) normal to groups: a at spacings of 2a and 3a. As mentioned 3+ 2+ nAl +3nSiO ? previously, Hoskins et al. (1989) suggested that n& +3(nÀ1)Sb3+ + 3SiOH3+ (4) the diffuse layers indicate that although there is an ordering effect in any one individual tunnel, it is Inanygivenchannel the amountof OH not strongly correlated with that of adjacent introduced by Sb-Si substitution is proportional tunnels. What form might this ordering take? to the number of &ÀSb chains, to a maximum of In end-member anhydrous dumortierite, formula one OH per Sb (where substitution 2 acts alone). Al7BSi3O18, all of the Al(1) sites are occupied by For very long &ÀSb chains (i.e. as n ? ?), the Al3+. Inholtite (andsome dumortierite samples) at amount of OHÀ introduced in this fashion is least some of the Al positions are occupied by negligible, and the average Sb-Si substitution is 5+ pentavalent cations, mainly Ta but sometimes 3+ 2+ 3+ Al + 3SiO ?&+ 3Sb (5) Nb5+ as well. The substitutionof pentavalent cations for Al3+ requires vacancies (&)atsome In total a number of vacancies equal to 1/3 of Al positions to maintain charge balance: (Sb + As) are introduced at the Al(1) site. Note 3+ 5+ that these vacancies must necessarily be at Al(1) 5Al ? 3Ta +2& (1) for coordination reasons. The number of vacancies introduced at the Al Summing expression 4 over all &ÀSb chains sites by this substitutionequals 2/3 the numberof inthe structure results inthe equation pentavalent cations at Al(1). 3+ 2+ nAl +3nSiO ? Independent of (Ta, Nb) content, when an Sb n& +3(nÀm)Sb3+ +3mSiOH3+ (6) site is occupied, the nearby O(2) or O(7) site must be vacant. As substitution of an Sb3+ (or As3+) ion where n is the total number of vacancies created at one Sb site obviates the octahedral coordination and m is the total number of &ÀSb chains, with
10 4 6 THE CRYSTAL CHEMISTRY OF HOLTITE
TABLE 6. Holtite site compositions from X-ray diffraction structure refinement and stoichiometric analysis.
H1 H2 H3 H4
Al(1) site Al 0.55 0.56 0.46 0.62 7.15 7.28 5.98 8.06 Ta 0.32 0.32 0.26 0.11 23.36 23.36 18.98 8.03 & 0.13 0.12 0.28 0.28
Al(2) site Al 1.97 1.99 1.99 2.00 25.61 25.74 25.87 26.00 & 0.03 0.01 0.02 0.00
Al(3) site Al 1.97 1.98 1.99 2.00 25.61 25.74 25.87 26.00 & 0.03 0.02 0.01 0.00
Al(4) site Al 1.83 1.86 1.94 1.93 23.79 24.18 25.22 25.09 & 0.17 0.14 0.06 0.07
Si, Sb sites Si 2.61 2.64 2.15 2.18 36.54 36.96 30.10 30.52 Sb + As 0.39 0.35 0.84 0.82 17.19 15.33 41.94 38.22
O 17.61 17.64 17.15 17.18 140.88 141.12 137.20 137.44 Total Al 6.33 6.39 6.37 6.55 & 0.36 0.29 0.37 0.35 & excl. Al1 0.23 0.17 0.09 0.07
Calculated* Total Al 6.34 6.35 6.29 6.55 Total & 0.34 0.33 0.45 0.34 & from Ta 0.21 0.22 0.17 0.07 & from Sb 0.13 0.12 0.28 0.27 O 17.61 17.65 17.16 17.18
Site populations are given in a.p.f.u. and e.p.f.u. (in italics).
* Calculated based on refined Ta and Sb+As and the formula Al7À˙Ù¯xÀˆÙ¯yTax&˜Ù¯x+ˆÙ¯yBSi3Ày(Sb,As)yO18Ày.
m 4 n. The OH groups are located at O(2) or transforming end-member dumortierite to holtite O(7). Substitution6 implies that anhydrousholtite thenwe have a generalholtite stoichiometry of has only a small number of very long &ÀSb Al7À[5x+y+z]/3(Ta,Nb)x&[2x+y+z]/3BSi3Ày chains, so that m << n and the equation reduces to (Sb,As) O (OH) (7) substitution5. y 18ÀyÀz z Giventhe above the formula of (Ta- andNb- where x is the total number of pentavalent cations, free) end-member holtite would be y is the total amount of Sb + As, and z is the total Al6B(Sb,As)3O15. The channels would be amount of OH, with z 4 y (z = y whenall Sb rings completely empty (vacancies at all Al(1), Si(1), are preceded and followed by Si rings). We ignore Si(2), O(2) and O(7) sites) which suggests that any di- and tetravalent cations substituting at the end-member holtite might be a useful molecular Al sites, tetrahedral substitutions at the Si or Sb sieve. sites, or additional anion substitutions. In parti- If we assume that the above Ta-Al and Sb-Si cular, we neglect the Al(1) vacancy substitution substitutions are the only substitutions involved in mechanism Al3+ + 3SiO2+ ?&+ 3SiOH3+
10 47 L. A. GROAT ET AL. proposed by Moore and Araki (1978) for The formula of holtite dumortierite, which corresponds to z 5 0, y =0 in formula 7. In addition, we have not considered Formula 7 with approximate values for x and y the possibility of tetrahedral Al, which is reported takenfrom the electronmicroprobe compositions innatural dumortierite, although rarely inexcess and maximum z (= y) results inthe following of 0.15 Al per formula unit (Grew, 1996). general formula for holtite I: Calculating holtite formulae assuming Al5.83(Ta,Nb)0.50&0.67BSi2.50(Sb,As)0.50 Si+Sb+As+P = 3 using the results of the micro- O (OH) probe analyses yields z < 0 for H4, and z ~ y for 17.00 0.50 H1, H2, H3, i.e. reasonable H2O contents for the and the following general formula for holtite II: latter three holtite samples, but a negative H2O Al5.92(Ta,Nb)0.25&0.83BSi2.00(Sb,As)1.00 content for H4. Assuming 0.122 Al in tetrahedral O (OH) coordination gives z = y = 0.919 for H4, which 16.00 1.00 corresponds to 1.28 wt.% H2O. Thus, the presence of tetrahedral Al is plausible inH4, but not in Dumortierite, holtite I and holtite II: distinct holtite from Greenbushes or Voron’i Tundry. species? Formula 7 with z = 0 (i.e. the anhydrous/long- chaincase) agrees well with the site populations Our analyses of holtite I and holtite II confirm the determined in the structure refinement, implying distinct clustering of compositions in terms of the only small amounts of OH are present at the O(2) Si/(Sb+As) ratio reported by Voloshin et al. (1987) and O(7) sites. Table 6 summarizes the octahe- and Voloshin and Pakhomovskiy (1988), although dral, total Si and Sb+As, and anion site the extremely limited number of localities for populations determined from structural refinement holtite (which effectively equates to a sampling and calculated from equation 6 using the refined limitation) may play a major role in the apparent Ta+Nb and Sb+As populations. Total Al and clustering. That holtite I and holtite II occur vacancy populations agree with calculated totals together at both the type locality and at Voron’i withinÔ0.08 a.p.f.u, with the largest discrepancy Tundry, with no obvious intermediate composi- for sample H3. The calculationconfirmsthat inall tions, could be interpreted to mean that these are four samples the number of vacancies at Al(1) is two different minerals separated by a miscibility equal to the number of vacancies due to Sb-Si gap. Alternatively, the clustering of compositions substitution (equation 5). The number of vacan- could be due to genetic rather than crystallographic cies created by Al-Ta substitutionat Al(1) factors. Inthe Voron’iTundry pegmatite, holtite I (equation1), however, are distributed amongthe is a early phase forming prisms 0.3À5cmlong, Al(2), Al(3) and Al(4) sites, with the greatest whereas holtite II is typically finely fibrous number at Al(4). (<1 mm) and later in the paragenetic sequence, In the absence of Sb-Si substitution, the replacing holtite I, stibiotantalite and microlite Al(1)ÀAl(1) distance is too short to allow (Voloshin et al., 1977, 1987; Voloshinand pentavalent cations at adjacent Al(1) positions, Pakhomovskiy, 1988). Differences in composition so Al(1) positions containing Ta5+ or Nb5+ must could have resulted from the different chemical be preceded and followed by vacant Al(1) sites. environments in which crystals of the two holtite This is likely to be the situation in pentavalent- varieties formed. Further calling into question the rich (or Ti4+-rich) dumortierite. However, in presence of immiscibility is the smaller composi- holtite the vacancies due to Sb-Si substitution tional gap between the two varieties in 390 may provide enough separation between penta- analyses of holtite II from Szklary, which give a valent cations to stabilize the channel without much greater range in the Si/(Sb+As) ratio introducing additional vacancies. (Pieczka, unpublished data). The crystal-structure The Al(4)O6 octahedronis both edge- andface- refinements do not indicate a fundamental sharing in its double chain. The cross-face difference in cation ordering that might serve as Al(4)ÀAl(4) distance is 2.58À2.61 A˚ inthe four a criterionfor recognizingholtite I andholtite II as samples, larger thanthe Al(1) ÀAl(1) distance but distinct species, and they would not qualify as still smaller thancross-edge Al ÀAl distances. distinct species using either the standard criterion This makes Al(4) the most favourable site for that at least one site must be occupied by a octahedral vacancies aside outside of the Al(1) differentcationor anionor the new criterionof the site. dominant-valency rule (Hatert and Burke, 2008).
10 4 8 THE CRYSTAL CHEMISTRY OF HOLTITE
As noted above, dumortierite may contain microbeam X-ray analysis of thick polished materi- significant Nb, Ta, As, Sb and Bi (Groat et al., als, thinfilms, andparticles . Microbeam Analysis, 4, 2001; Cempı´rek and Nova´k, 2004; Borghi et al., 177À200. 2004; Vaggelli et al., 2004). Are dumortierite and Borghi, A., Cossio, R., Fiora, L., Olmi, F. and Vaggelli, holtite separate species, or is holtite simply a Ta- G. (2004) Chemical determination of coloured zoned and Sb-rich variety of dumortierite? Our results minerals in ‘natural stones’ by EDS/WDS electron provide no evidence for a fundamental difference microprobe: anexample from dumortierite quartz- in formula, as Al is dominant at the Al(1) site and ites. X-ray Spectrometry, 33,21À27. Si is dominant at the Si(1) and Si(2) positions in Cempı´rek, J. and Nova´k, M. (2005) A greendumortier- both minerals. Moreover, available analyses ite from Kutna´ Hora, Moldanubicum, Czech suggest the possibility of a continuum in the Republic: spectroscopic and structural study. Pp. Si/(Sb + As) ratio betweenholtite I and 4À5in:International Meeting Crystallization Processes in Granitic Pegmatites (F. Pezzotta, dumortierite. editor). Elba, Italy, 2005, Book of Abstracts. In conclusion, we recommend that use of the Creagh, D.C. and Hubbell, J.H. (1992) International terms holtite I and holtite II be discontinued. A Tables for Crystallography,Vol C. Kluwer more precise definition of holtite and its Academic Publishers, Boston, USA, 200À206. distinction from dumortierite awaits detailed Creagh, D.C. and McAuley, W.J. (1992) International study of both minerals, including experiments to Tables for Crystallography,Vol C. Kluwer determine the maximum amount of As and Sb that Academic Publishers, Boston, USA, 219À222. canbe accommodated inthe dumortierite-holtite Cromer, D.T. and Waber, J.T. (1974) International structure. Such a definition will have to address, Tables for X-ray Crystallography,Vol. IV. The or at least consider, differences in diffraction Kynoch Press, Birmingham, UK. behaviour (i.e. presence or absence of super- Ferraris,G.,Ivaldi,G.andChopin,C.(1995) structure maxima), not just changes in bulk Magnesiodumortierite, a new mineral from very- chemistry. high-pressure rocks (WesternAlps). Part I: Crystal structure. European Journal of Mineralogy, 7, 167 174. Acknowledgements À Fuchs, Y., Ertl, A., Hughes, J.M., Prowatke, S., The authors thank R. Pogson of the Australian Brandsta¨tter, F. and Schuster, R. (2005) Museum and P.W. Powhat from the National Dumortierite from the Gfo¨hl unit: Lower Austria; Museum of Natural History (Smithsonian chemistry, structure, and infra-red spectroscopy. Institution) for samples from the type locality, European Journal of Mineralogy, 17, 173À183. N. Yonson for help with data collection at UBC, Golovastikov, N.I. (1965) The crystal structure of G. Poirier for help with data collectionat the dumortierite. Soviet Physics Doklady, 10, 493À495. Canadian Museum of Nature, R. Herbst-Irmer for Grew, E.S. (1996) Borosilicates (exclusive of tourma- help with twinning and E. Moffat for collecting line) and boron in rock-forming minerals in the infrared spectra. The manuscript was metamorphic environments. Pp. 387À502. in: improved by comments from an anonymous Boron: Mineralogy,Petrology,and Geochemistry. reviewer, Associate Editor E. Sokolova, and (E.S. Grew and L.M. Anovitz, editors). Reviews in Mineralogy, 33, Mineralogical Society of America, Editor M.D. Welch. Financial support was Chantilly, Virginia, USA. provided by the Natural Sciences and Grew, E.S., Graetsch, H., Po¨ter, B., Yates, M.G., Buick, Engineering Research Council of Canada in the I., Bernhardt, H.-J., Schreyer, W, Werding, G, form of a Discovery Grant to LAG. The Carson, C.J. and Clarke, G.L. (2008) Boralsilite, equipment in C-HORSE was purchased with the Al16B6Si2O37, and ‘boron-mullite’: compositional help of a grant from the Canadian Foundation for variations and associated phases in experiment and Innovation. nature. American Mineralogist, 93, 283À299. Groat, L.A., Grew, E.S., Ercit, T.S. and Piezcka, A. References (2001) The crystal chemistry of dumortierite and holtite, aluminoborosilicates with heavy elements. Alexander, V.D., Griffen, D.T. and Martin, T.J. (1986) Geological Society of America Abstracts with Crystal chemistry of some Fe- and Ti-poor Programs, 33, Abstract 383. dumortierites. American Mineralogist, 71, 786À794. Groat, L.A., Grew, E.S., Ercit, T.S. and Pieczka, A. Armstrong, J.T. (1995) CITZAF: A package of (2002) The crystal chemistry of holtite. Abstracts of correctionprograms for the quantitativeelectron the 18th General Meeting of the International
10 49 L. A. GROAT ET AL.
Mineralogical Association, p. 209. dumortierite. Mineralogical Magazine, 38,21À25. Hatert, F. and Burke, E.A.J. (2008) The IMAÀCNMNC Pryce, M.W. and Chester, J. (1978) Minerals of the dominant-constituent rule revisited and extended. Greenbushes Tinfield. Mineralogical Record, 9, The Canadian Mineralogist, 46, 717À728. 81À84. Hoskins, B.F., Mumme, W.G. and Pryce, M.W. (1989) Rossman, G.R. (1988) Vibrational spectroscopy of Holtite, (Si2.25Sb0.75)B[(Al6(Al0.43Ta0.27&0.30) hydrous components. Pp. 193À206. in: O15(O,OH)2.25]: crystal structure and crystal chem- Spectroscopic Methods in Mineralogy and Geology. istry. Mineralogical Magazine, 53, 457À463. (F.C. Hawthorne, editor). Reviews in Mineralogy, Ibers, J.A. and Hamilton, W.C. (1964) Dispersion 18, Mineralogical Society of America, Chantilly, corrections and crystal structure refinements. Acta Virginia, USA. Crystallographica, 17, 781À782. Vaggelli, G., Olmi, F., Massi, M., Giuntini, L., Fedi, M., Kazantsev, S.S., Pushcharovsky, D.Yu., Pasero, M., Fiora, L., Cossio, R. and Borghi, A. (2004) Chemical Merlino, S., Zubkova, N.V., Kabalov, Yu.K. and investigation of coloured minerals in natural stones Voloshin, A.V. (2005) Crystal structure of holtite I. of commercial interest. Microchimica Acta, 145, Crystallography Reports, 50,42À47. 249À254. Kazantsev, S.S., Zubkova, N.V. and Voloshin, A.V. Voloshin, A.V. and Pakhomovskiy, Ya.A. (1988) (2006) Refinement of composition and structure of Mineralogy of Tantalum and Niobium in Rare- holtite I. Crystallography Reports, 51, 412À413. Metal Pegmatites. Nauka, Leningrad (in Russian). Locock, A.J., Piilonen, P.C., Ercit, T.S. and Rowe, R. Voloshin, A.V., Gordienko, V.V., Gel’man, Ye.M., (2006) New mineral names. American Mineralogist, Zorina, M.L., Yelina, N.A., Kul’chitskaya, Ya.A., 91, 216À224. Men’shikov, Yu.P., Polezhayeva, L.I., Ryzhova, Moore, P.B. and Araki, T. (1978) Dumortierite, R.I., Sokolov, P.B. and Utochkina, G.I. (1977) Si3B[Al6.750.25O17.25(OH)0.75]: a detailed structure Holtite (first find in the USSR) and its relationship analysis. Neues Jahrbuch fu¨ r Mineralogie with other tantalum minerals in rare-metal peg- Abhandlungen, 132, 231À241. matites. Novyye Mineraly i Pervyye Nakhodki v Pekov, I.V. (1998) Minerals first discovered on the SSSR, 106(3), 337À347 (inRussian). territory of the former Soviet Union. Ocean Pictures, Voloshin, A.V., Pakhomovskiy, Ya.A. and Zalkind, Moscow. O.A. (1987) Aninvestigationof the chemical Pieczka, A. and Marszałek, M. (1996) Holtite À the first compositionandIR-spectroscopy of holtite. in: occurrence in Poland. Mineralogia Polonica, 27, Mineral’nyye Assotsiatsii i Mineraly 3À8. Magmaticheskikh Kompleksov Kol’skogo Platonov, A.N., Langer, K., Chopin, C., Andrut, M. and Polyostrova,Apatity,Kol’skiy Filial AN SSSR , Taran, M.N. (2000) Fe2+-Ti4+ charge-transfer in 14À34 (inRussian). dumortierite. European Journal of Mineralogy, 12, Werding, G. and Schreyer, W. (1990) Synthetic 521À528. dumortierite: its PTX-dependent compositional var- Pouchou, J.L. and Pichoir, F. (1985) ‘‘PAP’’ f(rZ) iations in the system Al2O3-B2O3-SiO2-H2O. procedure for improved quantitative microanalysis. Contributions to Mineralogy and Petrology, 105, Microbeam Analysis, 104À106. 11À24. Povarennykh, A.S. (1978) The use of infrared spectra for Zubkova, N.V., Pushcharovskii, D.Yu., Kabalov, Yu.K., the determination of minerals. American Kazantsev, S.S. and Voloshin, A.V. (2006) Crystal Mineralogist, 63, 956À959. structure of holtite II. Crystallography Reports, 51, Pryce, M.W. (1971) Holtite: a new mineral allied to 16À22.
1050